\documentclass[reqno]{amsart} \usepackage{hyperref} \AtBeginDocument{{\noindent\small \emph{Electronic Journal of Differential Equations}, Vol. 2017 (2017), No. 12, pp. 1--8.\newline ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu} \thanks{\copyright 2017 Texas State University.} \vspace{8mm}} \begin{document} \title[\hfilneg EJDE-2017/12\hfil An inverse spectral problem] {An inverse spectral problem for Sturm-Liouville operator with integral delay} \author[Manaf Dzh. Manafov \hfil EJDE-2017/12\hfilneg] {Manaf Dzh. Manafov} \dedicatory{In memory of M. G. Gasymov (1939-2008)} \address{Manaf Dzh. Manafov \newline Ad{\i}yaman University, Faculty of Science and Arts, Department of Mathematics, 02040, Ad{\i}yaman, Turkey} \email{mmanafov@adiyaman.edu.tr} \thanks{Submitted October 20, 2016. Published January 12, 2017.} \subjclass{34A55, 34L05, 47G20} \keywords{Sturm-Liouville Operator; inverse spectral problem; integral delay} \begin{abstract} In this article, we study an inverse spectral problem for Sturm-Liouville operator with integral delay. We prove that the standard spectral asymptotic conditions are necessary and sufficient for unique solvability of the inverse problem. \end{abstract} \maketitle \numberwithin{equation}{section} \newtheorem{theorem}{Theorem}[section] \allowdisplaybreaks \section{Introduction} We consider inverse problem for the boundary-value problem (BVP) generated by the integro-differential equation \begin{equation} l_y:=-y''+q(x)y+\int_0^{x}M(x-t)y(t)dt=\lambda ^{2}y,\quad x\in (0,a) \cup (a,\pi) \label{1.1} \end{equation} with the Dirichlet boundary conditions \begin{equation} U(y):=y(0)=0,\quad V(y):=y(\pi ) =0, \label{1.2} \end{equation} and the conditions at the point $x=a:$ \begin{equation} I(y):=\begin{cases} y(a+0)=y(a-0)\equiv y(a), \\ y'(a+0)-y'(a-0)=2\alpha \lambda y(a), \end{cases} \label{1.3} \end{equation} $q(x)$ and $M(x)$ are complex-valued functions, $q(x)\in L_2(0,\pi)$ and $(\pi -x) M(x)\in L_2(0,\pi )$, $\alpha \in \mathbb{C}$, $a\in (\frac{\pi }{2},\pi )$ and $\lambda$ is a spectral parameter. Sturm-Liouville spectral problems with potentials depending on the spectral parameter (in case $K(x)\equiv 0$) arise in various models of quantum and classical mechanics. For instance, the evolution equations that are used to model interactions between colliding relativistic spineless particles can be reduced to the form \eqref{1.1}. Then $\lambda ^{2}$ is associated with the energy of the system (see \cite{12,13}). Spectral problems of differential operators are studied in two main branches, namely, direct and inverse problems. Direct problems of spectral analysis consist in investigating the spectral properties of an operator. On the other hand, inverse problems aim at recovering operators from their spectral characteristics. Such problems often appear in mathematics, mechanics, physics, electronics, geophysics, meteorology and other branches of naturel sciences and engineering. Direct and inverse problems for the classical Sturm-Liouville operators have been extensively studied (see \cite{5,7,11} and the references therein). For integro-differential and other classes of nonlocal operators inverse problems are more difficult for investigation, and the classical methods either are not applicable to them or require essential modifications (see \cite{2,3,4,5,6,14,15}). In this aspect, various inverse spectral problems for the \eqref{1.1}, \eqref{1.3} BVP (special case $M(x)\equiv 0)$ have been investigated in \cite{8,9,10}). In this article we establish uniqueness result for inverse spectral problem for Sturm-Liouville operator with integral delay. \section{Integral representations for solutions} In this section, we construct an integral representation of the solution $y(x,\lambda )$ of \eqref{1.1}, \eqref{1.3}, satisfying the initial conditions \begin{equation} y(0,\lambda ) =1,\quad y'(0,\lambda )=i\lambda\,. \label{2.1} \end{equation} Also we study some properties of the solutions. Using the standard successive approximation methods (see \cite{11}), we can prove the following theorem. \begin{theorem} \label{thm2.1} The solution $y(x,\lambda )$ has the form \begin{equation} y(x,\lambda )=y_0(x,\lambda )+\int_{-x}^{x}A(x,t) e^{i\lambda t}dt, \label{2.2} \end{equation} where $y_0(x,\lambda )=\begin{cases} e^{ix\lambda }, & xa \end{cases}$ and the function $A(x,t)$ satisfies \begin{equation} \int_{-x}^{x}| A(x,t) | dt\leq e^{C\sigma _0(x)}-1 \label{2.3} \end{equation} with $\sigma _0(x)=\int_0^{x}(x-t) [ |q(t)| +\int_0^{t}| M(t-\tau ) |d\tau ] dt,$ and $C=1+2|\alpha|$. \end{theorem} \begin{proof} It is clear that when $\alpha =0$, if we consider the equation \eqref{1.1} separately on the intervals $(0,a)$ and $(a,\pi )$, we can write the solutions as \begin{gather} e_0(x,\lambda )=e^{ix\lambda }+\int_{-x}^{x}K_0(x,t) e^{i\lambda t}dt,\quad 0\leq xa, \label{2.5} \end{gather} respectively. For the solutions of the above equations to solve the equation that has representation \eqref{2.5}, the following equality must be satisfied: \begin{align*} &\int_{-x+2a}^{x}K_a(x,t) e^{i\lambda (t-a)}dt \\ &=\frac{1}{\lambda }\int_a^{x}\sin \lambda (x-t) \Big\{ q(t)\Big[e^{i\lambda (t-a)}+\int_{-t+2a}^{t}K_a(t,\tau ) e^{i\lambda (\tau -a) }d\tau \Big] \\ &\quad + \int_0^{t}M(t-\tau ) \Big[ e^{i\lambda (\tau -a)}+\int_{-\tau +2a}^{\tau }K_a(\tau ,s) e^{i\lambda ( s-a) }ds\Big] d\tau \Big\} dt\,. \end{align*} It is easy to obtain the integral equation \begin{equation} \label{2.6} \begin{aligned} K_a(x,t) &= \frac{1}{2}\int_a^{\frac{x+t}{2}}q(u)du +\frac{1 }{2}\int_a^{x}q(u)\int_{t-(x-u) }^{t+(x-u)}K_a(u,v) \,dv\,du \\ &\quad +\frac{1}{2}\int_a^{x}\int_{t-(x-u) }^{t+(x-u) }M(u-v) \,dx\,du \\ &\quad +\frac{1}{2}\int_a^{x}\int_0^{u}\int_{t-(x-u) }^{t+(x-u) }M(u-v) K_a(v,\xi ) d\xi \,dx\,du. \end{aligned} \end{equation} Since $e_a(x,-\lambda )$ is also the solution of \eqref{1.1}, \eqref{1.3} on the interval $0a$ and $A_{1}(x,t) =0$ for $|t| >x$. Now using the expression \eqref{2.10} in \eqref{2.9}, we have for $a0$ is a constant and $\sigma _a(x)=\int_a^{x}(x-t) \big[ q(t)+\int_0^{t}| M(t-\tau ) | d\tau \big] dt.$ Using \eqref{2.15}, from \eqref{2.11} and \eqref{2.13}, we have the estimate \begin{equation} \int_{-x}^{x}| A_2(x,t)| dt\leq e^{C\sigma _0(x)}-1 \label{2.16} \end{equation} for some constant $C>0$. Hence, from \eqref{2.14} and \eqref{2.16}, we arrive at \eqref{2.3}. \end{proof} Let $s(x,\lambda )$ be a solution of \eqref{1.1} with initial conditions $s(0,\lambda )=0,\quad s'(0,\lambda )=1.$ Because $y(x,\lambda )$ and $y(x,-\lambda )$ are two linearly independent solutions of \eqref{1.1}, \eqref{1.3}, then $s(x,\lambda )=\frac{y(x,\lambda )-y(x,-\lambda )}{2i}.$ Using integral representation \eqref{2.2}, we easily obtain \begin{equation} s(x,\lambda )=s_0(x,\lambda )+\int_0^{x}G(x,t)\frac{\sin \lambda t}{ \lambda }dt, \label{2.17} \end{equation} where $s_0(x,\lambda )=\begin{cases} \frac{\sin \lambda x}{\lambda }, & xa, \end{cases}$ $G(x,t)=A(x,t)-A(x,-t)$ is a continuous function, and $G(x,0)=0$. \section{Properties of the spectral characteristics} In the section, we study properties of eigenvalues and eigenfunctions of \eqref{1.1}. Let $y(x)$ and $z(x)$ be continuously differentiable functions on $(0,a)$ and $(a,\pi )$. Denote $\langle y,z\rangle :=yz'-y'z$. If $y(x)$ and $z(x)$ satisfy the matching conditions \eqref{1.3}, then \begin{equation} \langle y,z\rangle _{x=a-0}=\langle y,z\rangle_{x=a+0}, \label{3.1} \end{equation} i.e. the function $\langle y,z\rangle$ is continuous on $(0,\pi )$. Denote $\Delta (\lambda )=s(\pi ,\lambda )$. The eigenvalues $\{ \lambda _n^{2}\} _{n\geq 1}$ of the BVP \eqref{1.1} coincide with the zeros of the function $\Delta (\lambda )$. \begin{theorem} \label{thm3.1} The eigenvalues $\lambda _n^{2}$ and eigenfunctions $s(x,\lambda _n)$ of the BVP \eqref{1.1} satisfy the following asymptotic estimates for sufficiently large $n$, \begin{gather} \lambda _n=\lambda _n^{0}+o\big(\frac{1}{\lambda _n^{0}}\big) , \label{3.2} \\ s(x,\lambda _n) = o\big(\frac{1}{\lambda ^{0}}\big) +\begin{cases} \frac{\sin \lambda _n^{0}x}{\lambda _n^{0}} , & xa, \end{cases} \label{3.3} \end{gather} where $\lambda _n^{0}$ are the roots of $\Delta _0(\lambda ):=( 1-i\alpha ) \frac{\sin \lambda \pi }{\lambda }+i\alpha \frac{\sin \lambda (2a-\pi )}{\lambda }$ and $\lambda _n^{0}=n+h_n$, $h_n\in l_{\infty }$. \end{theorem} \begin{proof} From \eqref{2.17}, we have \begin{equation} \Delta (\lambda )=(1-i\alpha ) \frac{\sin \lambda \pi }{\lambda } +i\alpha \frac{\sin \lambda (2a-\pi )}{\lambda }+\int_0^{\pi }G(\pi ,t) \frac{\sin \lambda t}{\lambda }dt. \label{3.3.} \end{equation} Denote $\Gamma _n:=\{ \lambda :| \lambda | =\lambda _n^{0}+\delta \}$, $n=0,1,\dots,(\delta >0)$. Since $\Delta (\lambda )-\Delta _0(\lambda )=o(\frac{e^{| \operatorname{Im} \lambda | \pi }}{| \lambda | })$ and $| \Delta _0(\lambda )| \geq C_{\delta } \frac{e^{| \operatorname{Im}\lambda | \pi }}{| \lambda| }$ for all $\lambda \in \Gamma _n$, we establish by the Rouche's Theorem (see \cite[p. 125]{4c}) that $\lambda _n=\lambda _n^{0}+\varepsilon _n$, where $\varepsilon _n=o(1)$. Moreover, $\varepsilon _n=o(\frac{1}{\lambda _n^{0}})$ is obtained from the equality $o=\Delta (\lambda _n)=(\Delta _0'(\lambda _n^{0})+o(1)) \varepsilon _n +o(\frac{1}{\lambda _n^{0}})$. This completes the proof of \eqref{3.2}. From \eqref{2.17} and \eqref{3.2}, one can easily prove that the asymptotic formula \eqref{3.3} is true. \end{proof} \begin{theorem} \label{thm3.2} The specification of the spectrum $\{ \lambda_n^{2}\} _{n\geq 1}$ uniquely determines the characteristic function $\Delta (\lambda )$ by the formula \begin{equation} \Delta (\lambda )=[ (1-i\alpha ) \pi +i\alpha (2a-\pi )] \prod_{n=1}^{\infty }\frac{\lambda _n^{2}-\lambda ^{2}}{(\lambda _n^{0}) ^{2}}. \label{3.4} \end{equation} \end{theorem} \begin{proof} It follows from \eqref{3.3.} and consequently by Hadamard's factorization theorem (see \cite[p. 289]{4c}), $\Delta (\lambda )$ is uniquely determined up to a multiplicative constant by its zeros: \begin{equation} \Delta (\lambda )=C\prod_{n=1}^{\infty }(1-\frac{\lambda ^{2}}{ \lambda _n^{2}}) . \label{3.6} \end{equation} Consider the function \begin{align*} \Delta _0(\lambda ) &:=(1-i\alpha ) \frac{\sin \lambda \pi }{ \lambda }+i\alpha \frac{\sin \lambda (2a-\pi )}{\lambda } \\ &=[ (1-i\alpha ) \pi +i\alpha (2a-\pi )] \prod_{n=1}^{\infty}(1-\frac{\lambda ^{2}}{(\lambda _n^{0}) ^{2}}) . \end{align*} Then $\frac{\Delta (\lambda )}{\Delta _0(\lambda )} =C\frac{1}{[ (1-i\alpha ) \pi +i\alpha (2a-\pi )] } \prod_{n=1}^{\infty}\frac{(\lambda _n^{0}) ^{2}}{\lambda ^{2}} \prod_{n=1}^{\infty }\Big(1+\frac{\lambda _n^{2}-(\lambda_n^{0}) ^{2}} {(\lambda _n^{0}) ^{2}-\lambda ^{2}}\Big) .$ Taking \eqref{3.2} and \eqref{3.3.} into account we calculate $\lim_{\lambda \to -\infty }\frac{\Delta (\lambda )}{\Delta_0(\lambda )}=1,\quad \lim_{\lambda \to -\infty }\prod_{n=1}^{\infty }\Big(1+\frac{\lambda _n^{2}-(\lambda _n^{0}) ^{2}}{(\lambda _n^{0}) ^{2}-\lambda ^{2}} \Big) =1$ and hence $C=[ (1-i\alpha ) \pi +i\alpha (2a-\pi )] \prod_{n=1}^{\infty }\frac{\lambda _n^{2}}{(\lambda_n^{0}) ^{2}}.$ Substituting this into account \eqref{3.6} we arrive at \eqref{3.4}. \end{proof} \section{Formulation of the inverse problem uniqueness theorem} In this section, we study inverse problem of recovering $M(x)$ from the given spectral characteristics. We denote the BVP \eqref{1.1}-\eqref{1.3} by $L=L(M)$. Together with $L=L(M)$ we consider a BVP $\widetilde{L}=L(\widetilde{M})$ of the same form, but with different kernel $\widetilde{M}$. \smallskip \noindent\textbf{Inverse Problem:} Given a function $q(x)$, numbers $\alpha,a$, and the spectrum $\{ \lambda _n\} _{n\geq 1}$, construct the function $M(x)$. Let us prove the uniqueness theorem for the solution of the Inverse Problem. Everywhere below if a certain symbol $e$ denotes an object to $L$, then the corresponding symbol $\widetilde{e}$ denotes the analogous object related to $\widetilde{L}$ and $\widehat{e}=e-\widetilde{e}$. \begin{theorem} \label{teo 4.1}Fix $b\in (0,a)$. Let $\Lambda \subset \mathbb{N}$ be a subset of nonnegative integer numbers, and let $\Omega :=\{\lambda _n^{2}\} _{n\in \Lambda }$ be a part of the spectrum of $L$ such that the system of functions $\{ \cos \lambda _nx\} _{n\in \Lambda }$ is complete in $L_2(0,\pi )$. Let $M(x)=\widetilde{M}(x)$ almost everywhere (a.e.) on $(b,\pi )$, and $\Omega = \widetilde{\Omega }$. Then $M(x)=\widetilde{M}(x)$ a.e. on $(0,\pi)$. \end{theorem} \begin{proof} Let $\chi (x,\lambda )$ be the solution of the equation \begin{equation} l_{z}^{\ast }:=-z''+q(x)z+\int_{x}^{\pi }M(t-x) z(t)dt=\lambda ^{2}z,\quad x\in (0,a) \cup (a,\pi ) \label{4.1} \end{equation} under the conditions $\chi (\pi ,\lambda ) =0$, $\chi '(\pi ,\lambda ) =-1$ and the conditions at the point $x=a:\chi(a+0,\lambda ) =\chi (a-0,\lambda ) \equiv \chi(a,\lambda )$, $\chi '(a+0,\lambda ) -\chi '(a-0,\lambda ) =2\alpha \lambda \chi (a,\lambda)$. Denote $\Delta ^{\ast }(\lambda )=\chi (0,\lambda )$. Then by \eqref{3.1} we have \begin{align*} &\int_0^{\pi }\chi (x,\lambda ) \int_0^{x}\widehat{M}(x-t) \widetilde{s}(t,\lambda ) \,dt\,dx \\ &=\int_0^{\pi }\chi (x,\lambda ) l\widetilde{s}(x,\lambda )dx -\int_0^{\pi }\chi (x,\lambda ) \widetilde{l}\widetilde{s} (x,\lambda ) dx \\ &=\int_0^{\pi }l^{\ast }\chi (x,\lambda ) \widetilde{s}(x,\lambda ) dx -\int_0^{\pi }\chi (x,\lambda ) \widetilde{l}\widetilde{s}(x,\lambda ) dx \\ &\quad +[ \widetilde{s}(x,\lambda ) \chi '( x,\lambda ) -\widetilde{s}'(x,\lambda ) \chi(x,\lambda ) ] (| _0^{a} +|_a^{\pi }) \\ &=\Delta ^{\ast }(\lambda )-\widetilde{\Delta }(\lambda ) . \end{align*} For $\widetilde{l}=l$ we have $\Delta ^{\ast }(\lambda )\equiv \Delta (\lambda )$, and consequently \begin{equation} \int_0^{\pi }\chi (x,\lambda ) \int_0^{x}\widehat{M}(x-t) \widetilde{s}(t,\lambda ) \,dt\,dx=\widehat{\Delta }(\lambda ) . \label{4.2} \end{equation} We transform \eqref{4.2} into \begin{equation} \int_0^{\pi }\widehat{M}(x) \Big(\int_{x}^{\pi }\chi ( t,\lambda ) \widetilde{s}(t-x,\lambda ) dt\Big) dx = \widehat{\Delta }(\lambda ) . \label{4.3} \end{equation} Denote $w(x,\lambda ) =\chi (\pi -x,\lambda )$, $N(x)=M(\pi -x)$, \begin{equation} \varphi (x,\lambda ) =\int_0^{x}w(t,\lambda ) \widetilde{s}(x-t,\lambda ) dt. \label{4.4} \end{equation} Then \eqref{4.2} takes the form \begin{equation} \int_0^{\pi }\widehat{N}(x)\varphi (x,\lambda ) dx =\widehat{\Delta }(\lambda ) . \label{4.5} \end{equation} For $x\leq a$ the following representation holds \cite{14}, \begin{equation} \varphi (x,\lambda ) =\frac{1}{2\lambda ^{2}} \Big(-x\cos \lambda x+\int_0^{x}V(x,t) \cos \lambda t\,dt\Big) , \label{4.6} \end{equation} where $V(x,t)$ is a continuous function which does not depend on $\lambda$. Since $\Omega =\widetilde{\Omega }$, we have by Theorem \ref{thm3.2} $\Delta (\lambda )\equiv \widetilde{\Delta }(\lambda ) \Longrightarrow \widehat{\Delta }(\lambda ) \equiv 0.$ Then, substituting \eqref{4.6} into \eqref{4.5}, we obtain $\int_0^{b}\Big(-x\widehat{N}(x) +\int_{x}^{b}V( t,x) \widehat{N}(t) \Big) \cos \lambda x\,dx\equiv 0,$ and consequently, $-x\widehat{N}(x) +\int_{x}^{b}V(t,x) \widehat{N} (t) dt=0\quad \text{a.e. on }(0,b).$ Since this homogeneous Volterra integral equation has only the trivial solution it follows that $\widehat{N}(x) =0$ a.e. on $(0,b)$, i.e. $M(x)=\widetilde{M}(x)$ a.e. on $(0,\pi )$. \end{proof} \subsection*{Acknowledgements} The author wants to thank the anonymous referees for their valuable suggestions that improving this article. This work was supported by Grant No FEFMAP/2016-0002 from Adiyaman University of Research Project Coordination (ADYUBAP), Turkey. \begin{thebibliography}{99} \bibitem{2} Buterin, S. A.; On an inverse spectral problem for a convolution integro-differential operator, \textit{Result. Math.}, 50 (2007), 173-181. \bibitem{3} Buterin, S. A.; On the reconstruction of a convolution perturbation of the Sturm-Liouville operator from the spectrum, \textit{ Diff. Uravneniya}, 46:1 (2010), 146-149; English transl.: \textit{Diff. Equations}, 46:1 (2010), 150-154. \bibitem{4} Buterin, S. A.; Choque Rivero, A. E.; On inverse problem for a convolution integro-differential operator with Robin boundary conditions, \textit{Appl. Math. Letters}, 48 (2015), 150-155. \bibitem{4c} Conway, J. B.; \textit{Functions of One Complex Variable}. Springer-Verlag, New York, USA 2nd ed., 1995. \bibitem{5} Frelling, G.; Yurko, V.; \textit{Inverse Sturm-Liouville Problems and Their Applications}, Nova Science Publ., Inc: Huntington, NY, 2001. \bibitem{6} Kuryshova, Ju. V.; Inverse spectral problem for integro-differential operators, \textit{Math. Zametki}, 81:6 (2007), 855-866; English transl. \textit{Math. Notes}, 81:6 (2007), 767-777. \bibitem{7} Levitan, B. M.; \textit{Inverse Sturm-Liouville Problems}. VSP:Zeist, 1987. \bibitem{8} Manafov, M. Dzh.; Half-inverse spectral problem for differential pensils with interaction-point and eigenvalue-dependent boundary conditions. \textit{Hacettepe J. of Math. and Stats.}, 42:4 (2013), 339-345. \bibitem{9} Manafov, M. Dzh.; Kablan, A.; Inverse spectral and inverse nodal problems for energy-dependent Sturm-Liouville equations with $\delta$-interaction, \textit{Electronic J. of Diff. Equations}, vol. 2015 (2015), no. 26, 1-10. \bibitem{10} Manafov, M. Dzh.; Inverse spectral problems for energy-dependent Sturm-Liouville equations with finitely many point $\delta$-interactions, \textit{Electronic J. of Diff. Equations,} vol. 2016 (2016), no. 11, 1-12. \bibitem{11} Marchenko, V. A.; \textit{Sturm-Liouville Operators and Their Applications}. Operator Theory: Advanced and Application, Birkhauser: Basel, 22, 1986. \bibitem{12} Markus, A. S.; \textit{Introduction to the Spectral Theory of Polynomial Operator Pensils.} Shtinitsa, Kishinev, 1986; English transl., AMS, Providense, 1988. \bibitem{13} Jonas, P.; On the spectral theory of operators associated with perturbed Klein-Gordon and wave type equations. \textit{J. Operator Theory,} 29 (1993), 207-224. \bibitem{14} Yurko, V. A.; An inverse problem for integro-differential operators, \textit{Mat. Zametki}, 50:5 (1991), 134-146; English transl.: \textit{Math. Notes}, 50:5-6 (1991), 1188-1197. \bibitem{15} Yurko, V. A.; An inverse spectral problems for integro-differential operators, \textit{Far East J. Math. Sci.} 92:2 (2014), 247-261. \end{thebibliography} \end{document}